This week's issue is backed by RSA and ThreatModeler.
Hope you had a great weekend!
Well, there’s probably not much point in diving into anything security-related with the absolute Charlie Foxtrot meltdown that NVIDIA and the AI industry saw last week. RIP to everyone’s retirement accounts. 🫡
While I’m sure the markets will bounce back (they always have some kind of way), the Q4 2026 AI Implosion didn’t stop the Frontier AI labs from making cyber announcements. Here’s what came out just last week:
9 Nov - Anthropic launches Fable 5.2 / Mythos 5.2
10 Nov - Google expands Gemini 4 Carbon availability
12 Nov:
OpenAI announces Daybreak Defense Factory
OpenAI announces Daybreak Red Team
Anthropic publishes Anvil automated red-teaming
And the public cyber stocks in response to this? Largely unphased.

Every square is a cyber announcement from the Frontier AI labs or market-moving AI news
At least no more than the broader tech market.
The market knows that the AI labs can’t check their own cyber homework.
PARTNER
How many agents are running in your environment? Most organizations don’t know.
Find every agent. Control every action. Govern access.
RSA asked a bank how many agents ran in its environment. Officially: zero. Unofficially? 4,000. Maybe more.
Most organizations don’t know: 82% found AI agents in their environment that they didn’t know about in the past year.
RSA Agent ID discovers sanctioned and shadow agents, secures high-risk actions with human approval, and governs each with a named owner and continuous certification.
Stop identity risk from moving at machine speed.
Table of Contents
😎 Vibe Check
Click the options below to vote on whether you are a practitioner, founder, or investor. Leave a comment, and I'll anonymously feature the best takes in the next issue!
When it comes to Frontier AI, should the world accept that AI is both writing most of the code and then turning around to find vulnerabilities in it?
Last issue’s vibe check:
Which security category makes a resurgence in Q4 2026?
🟨🟨⬜️⬜️⬜️⬜️ Email security
🟩🟩🟩🟩🟩🟩 API security
🟨🟨🟨🟨🟨⬜️ AI SOC
🟨🟨🟨🟨⬜️⬜️ Security awareness training
🟨🟨🟨🟨⬜️⬜️ Other (tell me)
Last week’s results were interesting, and they're a bit of a tell of where people believe the focus needs to be in this AI world. While there are undoubtedly more software and vulnerabilities being created than ever before (hence the votes for technical domains), humans are still the main interface to AI and everything it accesses (hence the non-technical domains).
Also, the AI SOC is dead; long live the AI SOC (also, no one knows what an AI SOC is 🙃 ). The AI SOC isn’t real; it can’t hurt you. Meanwhile:

Some of the top comments from last week’s vibe check:
💬 “Who will tell our CEOs to be aware of espionage if we don’t have security awareness?”
💬 “None of the above in isolation - Q4 and into 2027 ends up being a unification to platform story!”
💬 “Enterprise AI agents use APIs to take action. API security has never been adequately addressed so now you’ve got agents, MCPs and tools leveraging APIs that may or may not be secure. Developers can now build new APIs faster than ever with coding assistants-those APIs are likely not secure. It’s a fast growing attack surface that many companies overlook in my opinion.”
💰 Market Summary
Private Markets
8 venture funding rounds announced, totaling $233.2M in disclosed amounts
9 M&A announcements, including 1 divestiture, across 5 countries and 7 categories
56% of M&A announcements were service companies
1 company announced layoffs
Public Markets
$2.5B in convertible notes announced
1 credit facility announced, totaling $1.5M
1 acquisition of a private cybersecurity company announced by a public buyer
1 company had an earnings report last week - $QLYS ( ▲ 4.53% )

» Trends, year-over-year charts, and every deal live on The Signal →
PARTNER
Need an Agentic Partner for Threat Modeling?
Meet ThreatModeler NexusAI
Somewhere, a perfectly good threat model is sitting on a Confluence page nobody has opened since March. ThreatModeler Nexus is built for how software gets made today, and its agentic partner, NexusAI, keeps your model moving with new code. It builds models from your diagrams, docs, and repos, checks connected sources for drift, and stages every update with its reasoning for your architects to approve.
No more threat models left waiting on abandoned pages. Discover how ThreatModeler Nexus can make a difference in your org today.
🏢 Funding
» Connect Claude, Cursor, or any MCP client to The Signal via the MCP for free. «
Akamai Technologies (US) · $2.5B convertible notes | secure networking platform, from Undisclosed Qualified Institutional Buyers (more)
Daylight Security (IL) · $80.0M Series B | automated MDR platform, led by Goldman Sachs Growth Equity (more)
Finite State (US) · $27.0M Series C | firmware layer cybersecurity platform, led by The O.H.I.O. Fund (more)
Quilr (formerly QuilrAI) (US) · $20.0M Series A | AI agent-based SaaS governance platform, led by Blumberg Capital (more)
Cyabra (IL) · $1.5M credit facility | social media disinformation protection, from Alpha Capital Anstalt (SEC Filing)
SEC filings may reflect partial or interim terms, and the final round numbers can change.
🤝 Mergers & Acquisitions
QinetiQ US (US) · US defense technology and services business of QinetiQ | sold by QinetiQ to Serco, $292.0M (more)
Pentio (JP) · cloud security integrator for IAM and PKI | acquisition by E-Guardian, $12.5M (more)
Hut Six Security (GB) · security awareness and training platform | acquisition by BoxPhish, undisclosed (more)
Legitec (ES) · data protection, compliance and cybersecurity consultancy | acquisition by Phenna Group, undisclosed (more)
Pillar Security (IL) · AI risk monitoring and mitigation platform | acquisition by Vanta, undisclosed (more)
Specular (US) · threat and vulnerability management | acquisition by Cogent Security, undisclosed (more)
Tests Assured (US) · quality assurance and security testing services for devices | acquisition by Quest Global, undisclosed (more)
❌ Layoffs
📚 Great Reads
Glasswing Overwhelms Maintainers with Underwhelming Findings - I'm a sucker for great data, and my friend Patrick Garrity over at VulnCheck always delivers. VulnCheck analyzed the outputs of Project Glasswing after seven months, and let's just say it's a clear indicator of hype vs. reality.
Who actually gets cheaper security from AI? - My friend Frank Wang talks about how not every company will pay the same rates in the AI world in terms of time, effort, and money, and how results may vary.
*Sponsored
🧪 Labs
“This one weird trick. Limit resets hate to see him coming.”
🫡 Signing Off
Have questions, comments, or feedback? Just reply back and let me know.
If you find this newsletter useful and know others who would, I'd really appreciate it if you'd forward it to them!
Mike P
P.S. Feel free to connect with me on LinkedIn.


